| Message ID | 20230524083153.2046084-7-s.hauer@pengutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Add perf support to the rockchip-dfi driver | expand |
Hi, On Wed, May 24, 2023 at 10:31:34AM +0200, Sascha Hauer wrote: > The DDR_MON counters are free running counters. These are resetted to 0 > when starting them over like currently done when reading the current > counter values. > > Resetting the counters becomes a problem with perf support we want to > add later, because perf needs counters that are not modified elsewhere. > > This patch removes resetting the counters and keeps them running > instead. That means we no longer use the absolute counter values but > instead compare them with the counter values we read last time. Not > stopping the counters also has the impact that they are running while > we are reading them. We cannot read multiple timers atomically, so > the values do not exactly fit together. The effect should be negligible > though as the time between two measurements is some orders of magnitude > bigger than the time we need to read multiple registers. > > Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> > --- Reviewed-by: Sebastian Reichel <sebastian.reichel@collabora.com> -- Sebastian > > Notes: > Changes since v4: > - rephrase commit message > - Drop unused variable > > drivers/devfreq/event/rockchip-dfi.c | 52 ++++++++++++++++------------ > 1 file changed, 30 insertions(+), 22 deletions(-) > > diff --git a/drivers/devfreq/event/rockchip-dfi.c b/drivers/devfreq/event/rockchip-dfi.c > index 680f629da64fc..126bb744645b6 100644 > --- a/drivers/devfreq/event/rockchip-dfi.c > +++ b/drivers/devfreq/event/rockchip-dfi.c > @@ -38,11 +38,15 @@ > #define DDRMON_CH1_COUNT_NUM 0x3c > #define DDRMON_CH1_DFI_ACCESS_NUM 0x40 > > -struct dmc_usage { > +struct dmc_count_channel { > u32 access; > u32 total; > }; > > +struct dmc_count { > + struct dmc_count_channel c[RK3399_DMC_NUM_CH]; > +}; > + > /* > * The dfi controller can monitor DDR load. It has an upper and lower threshold > * for the operating points. Whenever the usage leaves these bounds an event is > @@ -51,7 +55,7 @@ struct dmc_usage { > struct rockchip_dfi { > struct devfreq_event_dev *edev; > struct devfreq_event_desc desc; > - struct dmc_usage ch_usage[RK3399_DMC_NUM_CH]; > + struct dmc_count last_event_count; > struct device *dev; > void __iomem *regs; > struct regmap *regmap_pmu; > @@ -85,30 +89,18 @@ static void rockchip_dfi_stop_hardware_counter(struct devfreq_event_dev *edev) > writel_relaxed(SOFTWARE_DIS, dfi_regs + DDRMON_CTRL); > } > > -static int rockchip_dfi_get_busier_ch(struct devfreq_event_dev *edev) > +static void rockchip_dfi_read_counters(struct devfreq_event_dev *edev, struct dmc_count *count) > { > struct rockchip_dfi *dfi = devfreq_event_get_drvdata(edev); > - u32 tmp, max = 0; > - u32 i, busier_ch = 0; > + u32 i; > void __iomem *dfi_regs = dfi->regs; > > - rockchip_dfi_stop_hardware_counter(edev); > - > - /* Find out which channel is busier */ > for (i = 0; i < RK3399_DMC_NUM_CH; i++) { > - dfi->ch_usage[i].access = readl_relaxed(dfi_regs + > + count->c[i].access = readl_relaxed(dfi_regs + > DDRMON_CH0_DFI_ACCESS_NUM + i * 20); > - dfi->ch_usage[i].total = readl_relaxed(dfi_regs + > + count->c[i].total = readl_relaxed(dfi_regs + > DDRMON_CH0_COUNT_NUM + i * 20); > - tmp = dfi->ch_usage[i].access; > - if (tmp > max) { > - busier_ch = i; > - max = tmp; > - } > } > - rockchip_dfi_start_hardware_counter(edev); > - > - return busier_ch; > } > > static int rockchip_dfi_disable(struct devfreq_event_dev *edev) > @@ -145,12 +137,28 @@ static int rockchip_dfi_get_event(struct devfreq_event_dev *edev, > struct devfreq_event_data *edata) > { > struct rockchip_dfi *dfi = devfreq_event_get_drvdata(edev); > - int busier_ch; > + struct dmc_count count; > + struct dmc_count *last = &dfi->last_event_count; > + u32 access = 0, total = 0; > + int i; > + > + rockchip_dfi_read_counters(edev, &count); > + > + /* We can only report one channel, so find the busiest one */ > + for (i = 0; i < RK3399_DMC_NUM_CH; i++) { > + u32 a = count.c[i].access - last->c[i].access; > + u32 t = count.c[i].total - last->c[i].total; > + > + if (a > access) { > + access = a; > + total = t; > + } > + } > > - busier_ch = rockchip_dfi_get_busier_ch(edev); > + edata->load_count = access * 4; > + edata->total_count = total; > > - edata->load_count = dfi->ch_usage[busier_ch].access * 4; > - edata->total_count = dfi->ch_usage[busier_ch].total; > + dfi->last_event_count = count; > > return 0; > } > -- > 2.39.2 >
diff --git a/drivers/devfreq/event/rockchip-dfi.c b/drivers/devfreq/event/rockchip-dfi.c index 680f629da64fc..126bb744645b6 100644 --- a/drivers/devfreq/event/rockchip-dfi.c +++ b/drivers/devfreq/event/rockchip-dfi.c @@ -38,11 +38,15 @@ #define DDRMON_CH1_COUNT_NUM 0x3c #define DDRMON_CH1_DFI_ACCESS_NUM 0x40 -struct dmc_usage { +struct dmc_count_channel { u32 access; u32 total; }; +struct dmc_count { + struct dmc_count_channel c[RK3399_DMC_NUM_CH]; +}; + /* * The dfi controller can monitor DDR load. It has an upper and lower threshold * for the operating points. Whenever the usage leaves these bounds an event is @@ -51,7 +55,7 @@ struct dmc_usage { struct rockchip_dfi { struct devfreq_event_dev *edev; struct devfreq_event_desc desc; - struct dmc_usage ch_usage[RK3399_DMC_NUM_CH]; + struct dmc_count last_event_count; struct device *dev; void __iomem *regs; struct regmap *regmap_pmu; @@ -85,30 +89,18 @@ static void rockchip_dfi_stop_hardware_counter(struct devfreq_event_dev *edev) writel_relaxed(SOFTWARE_DIS, dfi_regs + DDRMON_CTRL); } -static int rockchip_dfi_get_busier_ch(struct devfreq_event_dev *edev) +static void rockchip_dfi_read_counters(struct devfreq_event_dev *edev, struct dmc_count *count) { struct rockchip_dfi *dfi = devfreq_event_get_drvdata(edev); - u32 tmp, max = 0; - u32 i, busier_ch = 0; + u32 i; void __iomem *dfi_regs = dfi->regs; - rockchip_dfi_stop_hardware_counter(edev); - - /* Find out which channel is busier */ for (i = 0; i < RK3399_DMC_NUM_CH; i++) { - dfi->ch_usage[i].access = readl_relaxed(dfi_regs + + count->c[i].access = readl_relaxed(dfi_regs + DDRMON_CH0_DFI_ACCESS_NUM + i * 20); - dfi->ch_usage[i].total = readl_relaxed(dfi_regs + + count->c[i].total = readl_relaxed(dfi_regs + DDRMON_CH0_COUNT_NUM + i * 20); - tmp = dfi->ch_usage[i].access; - if (tmp > max) { - busier_ch = i; - max = tmp; - } } - rockchip_dfi_start_hardware_counter(edev); - - return busier_ch; } static int rockchip_dfi_disable(struct devfreq_event_dev *edev) @@ -145,12 +137,28 @@ static int rockchip_dfi_get_event(struct devfreq_event_dev *edev, struct devfreq_event_data *edata) { struct rockchip_dfi *dfi = devfreq_event_get_drvdata(edev); - int busier_ch; + struct dmc_count count; + struct dmc_count *last = &dfi->last_event_count; + u32 access = 0, total = 0; + int i; + + rockchip_dfi_read_counters(edev, &count); + + /* We can only report one channel, so find the busiest one */ + for (i = 0; i < RK3399_DMC_NUM_CH; i++) { + u32 a = count.c[i].access - last->c[i].access; + u32 t = count.c[i].total - last->c[i].total; + + if (a > access) { + access = a; + total = t; + } + } - busier_ch = rockchip_dfi_get_busier_ch(edev); + edata->load_count = access * 4; + edata->total_count = total; - edata->load_count = dfi->ch_usage[busier_ch].access * 4; - edata->total_count = dfi->ch_usage[busier_ch].total; + dfi->last_event_count = count; return 0; }
The DDR_MON counters are free running counters. These are resetted to 0 when starting them over like currently done when reading the current counter values. Resetting the counters becomes a problem with perf support we want to add later, because perf needs counters that are not modified elsewhere. This patch removes resetting the counters and keeps them running instead. That means we no longer use the absolute counter values but instead compare them with the counter values we read last time. Not stopping the counters also has the impact that they are running while we are reading them. We cannot read multiple timers atomically, so the values do not exactly fit together. The effect should be negligible though as the time between two measurements is some orders of magnitude bigger than the time we need to read multiple registers. Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> --- Notes: Changes since v4: - rephrase commit message - Drop unused variable drivers/devfreq/event/rockchip-dfi.c | 52 ++++++++++++++++------------ 1 file changed, 30 insertions(+), 22 deletions(-)